deformation of disordered materials: a statistical physics problem? · mesoscopic description: «...

Deformation of disordered materials: a statistical physics problem?

Jean-Louis Barrat

Université Grenoble AlpesInstitut universitaire de France

Institut Laue Langevin, Theory Group

Acknowledgements

Kirsten Martens

Alexandre Nicolas

Ezequiel Ferrero

Elisabeth Agoritsas

Lydéric Bocquet

Chen LiuFrancesco Puosi

Kamran Karimi

Jörg Rottler

Vishwas Venkatesh

arXiv:1708.09194Review of Modern Physics 2018

Outline

• Flow of (soft) amorphous solids

• Elastoplastic models of flow

• Mean field analysis

• Second order transition -Avalanches

• First order transition –Strain localisation

• Transients - Creep

•Disordered elastic solids, far below/above any « glass transition temperature/density•Extremely diverse in terms of scales (nm-cm) and strength (100Pa-100GPa).•Still share common features in their deformation mechanisms

(Yield stress fluids/soft glasses)

Non-linear rheology and yield

stress:

Stress-strain curves at fixed rate

Herschel Bulkley equation

Plastic response of a foam (I. Cantat, O. Pitois, Phys. of fluids 2006)

Deformation at low T proceeds trough well identified plastic events or shear transformations (Argon and Kuo, 1976)

Plastic response of a simulated Lennard-Jones glass (Tanguy, Leonforte, JLB, EPJ E 2006)

Stress-strain curve at low strain rate, low temperature, small systems

Prototypical example : « T1 » events in foams

(Princen, 1981) (Dennin 2006)

Colloidal glass, confocal microscopy Science 318 (2007)

Events are shear transformations of Eshelby type

•Plastic instability in a very local region of the medium (irreversible) under the influence of the local stress.

Malandro, Lacks, PRL 1998•Instability involves typically a few tens of particles and small shear strains (1 to 10%)

•Surroundings respond essentially as an homogeneous elastic medium (incompressible). Quadrupolar symmetry of the response. Puosi, Rottler, JLB, PRE 2014

Events are shear transformations of Eshelby type

Eshelby transformation: an inclusion within an elastic material undergoes a spontaneous change of shape (eigenstrain): circular to elliptical.

In an homeogeneous, linear elastic solid, the Induced shear stress outside the inclusion is proportional to the inclusion transformation strain and to the Eshelby propagator (response to two force dipoles):

Events are shear transformations of Eshelby typeBest seen in experiments trough correlation patterns

Colloidal paste under simple shear(Jensen, Weitz, Spaepen, PRE 2014)

At large deformation, individual flow events interact and organize in the form of « avalanches » with a broad distribution of amplitudes.

« Barkhausen » type behavior, encountered in many condensed matter systems with disorder (magnetic domain walls, contact lines, vortices in superconductors, earthquakes, friction -> elastic line pinned by external potential, « depinning » problem)

Stationnary plastic flow regime:

• Statistical physics problem: elementary events identified, space time interactions and correlations between events lead to avalanches and noise.

• Jamming/Yielding (from flowing suspension to a solid paste) has features of a dynamical phase transition (depinning ?)

• Universal features from granular media to metallic glasses

• Different (simpler ?) from extensively studied crystalline plasticity which involves topological flow defects (dislocations).

=> Gain qualitative understanding of phenomena at different scales with simple models, and tune model parameters for quantitative predictions.

Flow of amorphous solids:

• Microscopic : Particle based, molecular dynamics or athermal quasistatic deformations. Detailed information, limited sizes /times.

• Mesoscopic : Coarse grain and use the « shear transformations » as elementary events, with elastic interactions between them.

• Continuum : Stress, strain rate, and other state variables (« effective temperature ») treated as continuum fields.

Three levels of modelling

Outline

• Flow of soft amorphous solids






Linear elastic response at low shearRelated models by Bulatov and Argon, Picard et al

Mesoscopic elastoplastic model: « Ising model » of amorphous deformation:

At some point (yield criterion), the material yields locally and the energy stored locally is dissipated → plastic event

Related models by Bulatov and Argon, Picard et al

∂ t σ ∝−σ

Stress is redistributed during plastic event, owing to the presence of an elastic medium

Then elastic regime again, etc.

Implementation on a grid ; σi local stress variable

UniversalUniversal

Non universalNon universal

Mesoscopic description: « Ising model » of plastic deformation, implemented on a lattice; σi local stress variable

• Can be implemented in a finite element code rather than assuming Eshelby propagator. Allows for disorder/weakening (K. Karimi, LJB, PRE 2016) or various boundary conditions (Budrikis, Zapperi, Nat. Comm. 2017)

• Fully tensorial description and taking convection into account are possible (A. Nicolas, JLB, PRL2014). Validated against experiments (Goyon et al, flow in microchannels).

• Parameters can be related to microscopic description (Patinet, Falk, Vandembroucq ; Tanguy, Albaret)

Outline







Mean field analysis (Hébraud Lequeux model): Stress diffusion due to mechanical noise + self consistency

External drive Yield if σ>σcStress diffusion Reset to zero after yield

P(σ,t) probability distribution of stress on a typical site (no disorder, single local yield stress)

Non linear feedback

Yield rule and plastic activity

Mean field analysis (Hébraud Lequeux model): Stress diffusion due to mechanical noise + self consistency

A spatial version (inhomogeneous stress or strain rate) can be derived (Bocquet et al, PRL 2012) and compares well with experiments in confined geometries (A. Nicolas, JLB, PRL2014)

Disorder does not modify exponents (Agoritsas et al, 2016)

Mathematical study: Julien Olivier, Z. Angew. Math. Phys., 61(3), 2010, pp 445-466

Outline







Dynamical transition with “Second order” critical behaviour, monotonous flow curve. Avalanche behavior at vanishing strain rates, analogy with depinning problems.

Yield (arrested->flow) transition as a second order dynamical transition

Rewrite Herschel Bulkley equation as:

Similarity between the yielding transition to depinning ?

Lin, Lerner, Rosso, Wyart, EPL 2014, PNAS 2014

Major difference with depinning problem: Kernel G(r-r’) is not positive everywhereNew universality class(es), different from depinning problems

Avalanches: stress drops in the stress/strain curve.Transition characterized by avalanche statistics (similar to Barkhausen noise, earthquakes) obtained from numerics/experiments.

Lin et al, PNAS2014

Avalanche sizes display universal power laws in the limit of small strain rate (quasistatic)

« Earthquake like » statistics of stress drops

10-6

10-4

10-2

S/Ldf

10-3

100

103

LdftP

S

10-6

10-4

10-2

S/Ldf

10-3

100

103

LdftP

S

10-4

10-2

100

102

S

10-6

10-3

100

103

PS

10-4

10-2

100

102

S

10-6

10-3

100

103

PS

(a) (b)S-1.25

S-1.28

(d=3) (d=2)

Avalanche sizes display universal power laws in the limit of small strain rate (quasistatic)

Universal critical exponents, different from mean field

No complete theory of critical exponents yet (in contrast to depinning)

Outline







Yield (arrested->flow) transition as a first order dynamical transition: strain localisation / shear banding

Coexistence of flowing regions and solid regions at the same value of the stress

Granular pastes (Barentin et al., 2003)

Bubble Rafts (Dennin et al., 2004)

Chocolate (Coussot et al.)

Lennard-Jones glass (Simulation, Varnik, Bocquet, JLB, 2004)

« Explained » by static vs dynamic yield stress

4.5 5.0 5.5 6.0

0.000

0.002

0.004

4.75 5.00 5.25 5.50

0.0

2.0x10-4

4.0x10-4

Vel

ocity

(m

/s)

Distance (cm)

Ve

loci

ty (

m/s

)

Distance (cm)

Strain localisation/ Shear banding

Divoux, Fardin, Manneville, Lerouge, Annual review fluid mechanics 2016

Flow profile (cylindrical Couette)

Flow curve

Example system


A possible mechanism from a mesoscopic viewpoint (Coussot and Ovarlez, Martens et al): long plastic events (large “healing time”)

Coussot and Ovarlez mean field analysis (EPJE 2010)

Constitutive curve becomes non monotonic at large tres



Life cycle of a single block Flow curves

K Martens, L. Bocquet, JLB, Soft Matter 2012

Assembly of elastoplastic blocks interacting via elastic propagator.Healing time tres before elastic recovery varies.



Cumulated plastic activity

Martens, . Bocquet, JLB, Soft Matter 2012Tyukodi, Patinet, Roux, Vandembroucq 2016 “soft modes in the depinning transition”

Elastic propagator replaced by short range interaction

Why linear structure ?

=0

Outside an homogeneous plastic band (soft mode of the elastic propagator)


Other possibilities for shear banding :

- ageing (microscopic view, Shi and Falk, PRL 2007; mesoscopic, Vandembroucq and Roux, PRB 2011)

- Inertia (Nicolas et al, PRL 2016)

Strain localisation takes place in aged systems (from Vandembroucq and Roux)

Age increases

Outline







Siegenbürger et al, PRL 2012 Creep in a colloidal glass

Strain response: different stress levels, different waiting times Fluidization time behaves as a

power law of the distance to yield stress

Divoux et al, Soft Matter 2011Carbopol microgel

Creep: Apply a fixed stress σ and measure the strain g (t)

Stress controlled version of elastoplastic modelsChen Liu, Kirsten Martens, JLB Mean field version: arxiv:1705.06912; Phys rev lett. 2017Spatial version: arXiv:1807.02497 Soft matter 2018

Results qualitatively similar to experiments; very strong dependence on the initial condition for the probability distribution function

Sd: decreases when system ages

Fluidization time follows a power law with the static yield stress as a reference (can be identified with overshoot in stress strain curve).

Exponent is not universal, depends on system age.

In spatially resolved models, fluidisation is announced by a strong cooperativity in plastic activity – precursor of rupture (see arXiv:1807.02497 )

Conclusions

• Flow of soft solids can be described by simple elasto plastic models.

• Second « order » dynamical phase transition with similarities to depinning (avalanches, critical exponents) but different universality class.

• Possibility of first order transition (strain localisation…), several different mechanisms

• Non universal creep behaviour

• Perspectives: living tissues, temperature effects, critical behaviour…

Review paper : Review of Modern Physics 2018

https://statphys27.df.uba.ar/https://yielding2019.sciencesconf.org/

A Satellite meeting of :

Buenos Aires, 8th-12th July, 2019

https://statphys27.df.uba.ar/

https://statphys27.df.uba.ar/

https://yielding2019.sciencesconf.org/

https://yielding2019.sciencesconf.org/

Note: mechanical noise is different from thermal noise! (different from ‘’soft glassy rheology’’ picture)

A. Nicolas, K. Martens, JLB, EPL 2014E. Agoritsas et al, EPJ E 2015

• Thermal noise acts on strain variable l in a fixed landscape biased by the stress

• Mechanical noise acts a diffusive process on the stress bias itself

=> Very different escape times (Arrhenius vs diffusive)

Potential Energy Landscape Picture for a small region (STZ):

Understanding gained from studies of elastoplastic models:

•Mean field analysis: Hébraud-Lequeux 1998, Herschel Bulkley law with exponent 1/2

•Acceleration of diffusion by active deformation (Martens, Bocquet, JLB, PRL 2011)

•Interplay between aging and strain localisation (Vandembroucq and Roux, PRB 2011)

•criteria for strain localisation in soft materials (Martens, Bocquet, JLB, Soft Matter 2012)

•Introducing thermal activation of local events allows one to reproduce « compressed exponential » relaxation (Ferrero, Martens, JLB, PRL 2014).

As quenched

Aged

Accumulated plastic deformation (Roux Vandembroucq)

Phys Rev A, 1991

Spring network with threshold in force

Slope -1.4 in 2D

Mesoscopic description- an old idea

Model can reproduce quantitatively average flow and fluctuations of a dense suspension in microchannel geometries (experiments: Goyon et al, 2011; Nicolas and Barrat, PRL 2013) – Parameters calibrated on homogeneous flow curve.

Herschel Bulkley equation

Velocity profile Velocity fluctuations

Validation of the approach

• Statistical physics problem: elementary events identified, space time interactions and correlations between events lead to avalanches and noise.

• Jamming/Yielding (from flowing suspension to a solid paste) has features of a dynamical phase transition (depinning ?)

• Universal features from granular media to metallic glasses

• Different (simpler ?) from extensively studied crystalline plasticity which involves topological flow defects (dislocations).

=> Gain qualitative understanding of phenomena at different scales with simple models, and tune model parameters for quantitative predictions.

Flow of amorphous solids:

deformation of disordered materials: a statistical physics problem? · mesoscopic description: «...

Documents